CN111382862B - Method for identifying abnormal data of power system - Google Patents
Method for identifying abnormal data of power system Download PDFInfo
- Publication number
- CN111382862B CN111382862B CN201811609951.3A CN201811609951A CN111382862B CN 111382862 B CN111382862 B CN 111382862B CN 201811609951 A CN201811609951 A CN 201811609951A CN 111382862 B CN111382862 B CN 111382862B
- Authority
- CN
- China
- Prior art keywords
- neural network
- data
- power system
- training
- abnormal data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 30
- 230000002159 abnormal effect Effects 0.000 title claims abstract description 29
- 238000013528 artificial neural network Methods 0.000 claims abstract description 48
- 239000002245 particle Substances 0.000 claims abstract description 39
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 30
- 238000012549 training Methods 0.000 claims abstract description 29
- 230000000739 chaotic effect Effects 0.000 claims abstract description 15
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 claims abstract description 7
- 238000005070 sampling Methods 0.000 claims abstract description 6
- 239000011159 matrix material Substances 0.000 claims description 23
- 238000005259 measurement Methods 0.000 claims description 23
- 210000002569 neuron Anatomy 0.000 claims description 23
- 238000012360 testing method Methods 0.000 claims description 8
- 238000013016 damping Methods 0.000 claims description 4
- 101100348854 Aspergillus sp. (strain MF297-2) notE gene Proteins 0.000 claims description 3
- 230000004913 activation Effects 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 2
- 239000013307 optical fiber Substances 0.000 claims 1
- 238000004364 calculation method Methods 0.000 description 4
- 238000007781 pre-processing Methods 0.000 description 3
- 238000007619 statistical method Methods 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Economics (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Tourism & Hospitality (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Probability & Statistics with Applications (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Evolutionary Biology (AREA)
- General Business, Economics & Management (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention relates to a method for identifying abnormal data of an electric power system, which comprises the steps of taking normal data of the electric power system as a training sample, and training a neural network; inputting data to be detected into the trained neural network to obtain a residual sequence; clustering residual training based on an affine propagation clustering algorithm; and judging abnormal data according to the characteristics of each category and the intra-category distance. The invention utilizes the chaotic particle swarm algorithm to train the neural network, and simultaneously adopts the affine propagation clustering algorithm to realize data clustering, thereby obviously reducing the calculated amount, simultaneously not depending on sampling distribution, and effectively improving the accuracy of identifying abnormal data of the power system.
Description
Technical Field
The invention relates to the field of abnormal data identification, in particular to an abnormal data identification method for an electric power system.
Background
The detection and identification of abnormal data of the power system are one of the important functions of the state estimation of the power system, and the purpose of the method is to eliminate a small amount of accidental bad data in the measurement sampling data and improve the reliability of the state estimation. Complex power networks contain a large amount of real-time data, and the accuracy of the data determines the safety and reliability of the operation of the power system. Bad data in the power system may affect the dispatcher to make wrong decisions, thereby affecting the normal operation of the power system and even possibly threatening the safety of the whole power system.
Therefore, in order to ensure stable and safe operation of the power system, it is important to detect the bad data and extract them from the original data for correction. The quality of the measured data of the power system is an important factor influencing the state estimation efficiency and result of the power system, a small amount of bad data objectively exists in the measured data, and the detection and identification of the bad data are important components of the state estimation of the power system. The presence of bad data in the power system may degrade the convergence performance of the state estimation and may even cause less failure of the state estimation. How to reliably detect and correct bad data becomes a challenge for state estimation applications.
The method for identifying abnormal data of the power system generally comprises the steps of data preprocessing, clustering and abnormal data judgment. The BP neural network commonly adopted in the current data preprocessing method is easy to fall into a local extremum so as to reduce the data preprocessing precision. The clustering algorithm often employs a K-means method and a median-gap statistical method. However, the K-means method requires the number of clusters to be preset, and if the number of clusters is not properly set, the accuracy of the algorithm is seriously affected. Although the middle gap statistical method can automatically calculate the number of clusters, the algorithm depends on the sampling distribution of data, and a large amount of samples are needed to obtain a stable result, so that the algorithm has the defects of complex calculation and large calculation amount, and is not suitable for processing mass data.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides an abnormal data identification method for an electric power system, which solves the problems of complex calculation and large calculation amount of the existing abnormal data identification algorithm.
The technical scheme adopted by the invention for realizing the purpose is as follows:
a method for identifying abnormal data of an electric power system comprises the following steps:
step 1: training a neural network based on normal data of the power system;
step 1-1: acquiring a normal electrical appliance operation process state measurement value data set X:
each row in X is a measurement value of the operating process state of the electrical appliance, such as a voltage value, a current value, active power, reactive power and the like. In the matrix, there are a total of M measurements, each with N samples.
Step 1-2: normalizing X:
wherein xjminAnd xjmaxThe maximum value and the minimum value of the j measurement value are obtained.
Step 1-3: randomly splitting the data set X along the row direction to form a set XM×LAnd XM×(N-L)And L, N-L is approximately equal to 0.8:0.2 and is respectively used as a training set and a test set of the neural network.
Step 1-4: and (5) taking the measured value in the ith as a target value and the measured values of other types as input values, and training the neural network. The neural network is selected as a 3-layer neural network and comprises an input layer, a hidden layer and an output layer, the number of neurons of the input layer is M-1, the number of neurons of the hidden layer is set to be M-1, and the number of neurons of the output layer is 1. The relationship between neural network inputs and outputs is:
where f and g are activation functions, xiIs the input to the ith input layer neuron, αijIs the weight between the ith neuron of the input layer and the jth neuron of the hidden layer, βjIs the weight between the jth neuron of the hidden layer to the output node, AjAnd B is a bias coefficient and is initialized by adopting an Nguyen-Widrow algorithm. The error between the neural network output and the target value adopts the mean square error.
Step 1-5: and training the neural network based on the particle swarm algorithm of the chaotic search.
Step 1-5-1: velocity v of randomly initialized particlesiAnd position li;
Step 1-5-2: calculating a fitness value f for each particleiI.e. neural network outputs and targetsMean square error between values.
Step 1-5-3: obtaining the historical optimal position of each particle and the optimal positions of all the particles
1-5-4: updating the speed and the position according to a formula;
vid←w*vid+c1r1(Lid-xid)+c2r2(Lgd-xid)
lid←lid+vid
wherein: i is the particle index; d is the data dimension index, c1And c2Is a constant number r1And r2Is a closed interval [0,1]Random number of (3), LiAnd LgThe historical optimal position of the ith particle and the optimal positions of all particles, and w is the inertial weight.
1-5-5: if the iteration maximum times are not reached, returning to the step 2; if the maximum number of iterations is reached, the following steps are continued.
1-5-6: performing chaotic local search according to a formula; if the searched solution is more optimal, updating the historical optimal position;
ωi←μωi(1-ωi)
li←ai+di×ωi
wherein a isi,diIs constant, μ is the attractor.
And training the weight and the bias coefficient of the neural network on the training set by using a chaotic particle swarm algorithm, and testing on the training set until the mean square error between the output of the neural network on the testing set and a target value is not reduced.
Step 2: inputting data to be detected into the trained neural network to obtain a residual sequence;
step 2-1: acquiring a data set Y to be tested:
y is the same as the data type of each row in the normal data set X, but there are Q samples per measurement.
Step 2-2: normalizing X:
wherein xjminAnd xjmaxThe maximum value and the minimum value of the j measurement value in the training set.
Step 2-3: obtaining mean square error sequence E ═ E by using trained neural network1,e2,...,eQ]。
And step 3: clustering residual training based on an affine propagation clustering algorithm;
step 3-1: and randomly initializing an attraction degree matrix R, an attribution degree matrix A and a damping factor lambda between 0 and 1.
Updating the attraction matrix R and the attribution matrix A
D=R+A
Wherein s (i, k) ═ x | |i-xk||2D is the decision matrix for similarity.
Step 3-2: and finding out the point of the decision matrix D with the diagonal larger than 0 as a clustering center. And classifying the sample points according to the distance between the sample points and the clustering representative points.
Step 3-3: and if the iteration times are larger than the initialized maximum iteration times, stopping clustering. Otherwise, returning to the step 2.
And 4, step 4: and judging abnormal data according to the characteristics of each category and the intra-category distance.
Step 4-1: for a category with the number of elements less than or equal to 3, all elements in the category are considered to be abnormal data.
Step 4-2: for classes with element numbers greater than 3, the mean square error e for a certain inputiLet its corresponding cluster center beIf it is notE is theniAbnormal values exist in some dimensions of the corresponding measurement data vector.
The invention has the following beneficial effects and advantages:
the chaotic particle swarm algorithm is used for neural network training, and the affine propagation clustering algorithm is used for realizing data clustering, so that the calculated amount can be obviously reduced, the sampling distribution is not depended on, and the accuracy of abnormal data identification of the power system is effectively improved.
Drawings
FIG. 1 is a flow chart of a method of the present invention;
fig. 2 is a schematic diagram of the neural network structure of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in detail below. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein, but rather should be construed as modified in the spirit and scope of the present invention as set forth in the appended claims.
It will be understood that when an element is referred to as being "disposed on" another element, it can be directly on the other element or intervening elements may also be present. When an element is referred to as being "connected" to another element, it can be directly connected to the other element or intervening elements may also be present. The terms "front," "back," "left," "right," and the like as used herein are for illustrative purposes only and do not represent a unique embodiment.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The terminology used in the description of the invention herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.
Fig. 1 shows a flow chart of the method of the present invention.
The method comprises the following steps:
acquiring a normal electrical appliance operation process state measurement value data set X:
each row in X is a measurement value of the operating process state of the electrical appliance, such as a voltage value, a current value, active power, reactive power and the like. In the matrix, there are a total of M measurements, each with N samples.
Normalizing X:
wherein xjminAnd xjmaxThe maximum value and the minimum value of the j measurement value are obtained.
Randomly splitting the data set X along the row direction to form a set XM×LAnd XM×(N-L)And L, N-L is approximately equal to 0.8:0.2 and is respectively used as a training set and a test set of the neural network.
And (5) taking the measured value in the ith as a target value and the measured values of other types as input values, and training the neural network.
The neural network is selected as a 3-layer neural network and comprises an input layer, a hidden layer and an output layer, the number of neurons of the input layer is M-1, the number of neurons of the hidden layer is set to be M-1, and the number of neurons of the output layer is 1.
The relationship between neural network inputs and outputs is:
where f and g are activation functions, xiIs the input to the ith input layer neuron, αijIs the weight between the ith neuron of the input layer and the jth neuron of the hidden layer, βjIs the weight between the jth neuron of the hidden layer to the output node, AjAnd B is a bias coefficient and is initialized by adopting an Nguyen-Widrow algorithm. The error between the neural network output and the target value adopts the mean square error.
The network is generally trained by using intelligent algorithms such as a gradient back propagation algorithm or particle swarm. But convergence to a local minimum occurs. In order to solve the problem, a particle swarm algorithm based on chaotic search is adopted to train the neural network.
The particle swarm optimization has the core idea that the parameter vector to be optimized is regarded as the position of a particle, the objective function to be optimized is regarded as the fitness f of the particle, and the fitness is optimized by dynamically adjusting the position between the group history optimal position and the individual history optimal position. The basic particle swarm algorithm has the problem of convergence to a local optimal solution, so that the chaotic search is utilized to enhance the search capability of the particle swarm algorithm.
The complete chaotic particle swarm algorithm comprises the following steps:
step 1: velocity v of randomly initialized particlesiAnd position li;
Step 2: calculating a fitness value f for each particleiI.e. the mean square error between the neural network output and the target value.
And step 3: obtaining the historical optimal position of each particle and the optimal positions of all the particles
And 4, step 4: updating the speed and the position according to a formula;
vid←w*vid+c1r1(Lid-xid)+c2r2(Lgd-xid)
lid←lid+vid
wherein: i is the particle index; d is the data dimension index, c1And c2Is a constant number r1And r2Is a closed interval [0,1]Random number of (3), LiAnd LgThe historical optimal position of the ith particle and the optimal positions of all particles, and w is the inertial weight.
And 5: if the iteration maximum times are not reached, returning to the step 2; if the maximum number of iterations is reached, the following steps are continued.
Step 6: performing chaotic local search according to a formula; if the searched solution is more optimal, updating the historical optimal position;
ωi←μωi(1-ωi)
li←ai+di×ωi
wherein a isi,diIs constant, μ is the attractor.
And training the weight and the bias coefficient of the neural network on the training set by using a chaotic particle swarm algorithm, and testing on the training set until the mean square error between the output of the neural network on the testing set and a target value is not reduced.
Acquiring a data set Y to be tested:
y is the same as the data type of each row in the normal data set X, but there are Q samples per measurement.
Normalizing X:
wherein xjminAnd xjmaxThe maximum value and the minimum value of the j measurement value in the training set.
Obtaining mean square error sequence E ═ E by using trained neural network1,e2,...,eQ]。
Clustering the elements in the E based on an affine propagation clustering algorithm:
and randomly initializing an attraction degree matrix R, an attribution degree matrix A and a damping factor lambda between 0 and 1.
Updating the attraction matrix R and the attribution matrix A
D=R+A
Wherein s (i, k) ═ x | |i-xk||2D is the decision matrix for similarity.
And step 3: and finding out the point of the decision matrix D with the diagonal larger than 0 as a clustering center. And classifying the sample points according to the distance between the sample points and the clustering representative points.
And 4, step 4: and if the iteration times are larger than the initialized maximum iteration times, stopping clustering. Otherwise, returning to the step 2.
For a category with the number of elements less than or equal to 3, all elements in the category are considered to be abnormal data.
For classes with element numbers greater than 3, the mean square error e for a certain inputiLet its corresponding cluster center beIf it is notE is theniAbnormal values exist in some dimensions of the corresponding measurement data vector.
Table 1 shows the accuracy of determining abnormal data with different dimensions, and the number of data samples is 1000 points.
TABLE 1 abnormal data determination accuracy for different dimensions
Data dimension | 3 | 5 | 7 | 9 | 10 |
BP neural network-based and mesoscopic statistical method | 90.2% | 91.4% | 93.1% | 93.6 | 94.2% |
Proposing an algorithm | 94.8% | 95.2% | 96.6% | 97.2% | 98.5% |
As can be seen from table 1, the proposed algorithm has significant advantages in terms of accuracy over other algorithms.
Claims (7)
1. A method for identifying abnormal data of an electric power system is characterized by comprising the following steps:
step 1: taking normal data of the power system as a training sample to train a neural network;
step 2: inputting data to be detected into the trained neural network to obtain a residual sequence;
the step of inputting the data to be detected into the trained neural network to obtain the residual sequence comprises the following steps:
step 2.1: acquiring a data set Y to be tested:
the data type of each row in the data set Y to be tested is the same as that of each row in the normal data set X, but each measured value has Q sampling data;
step 2.2: normalizing Y:
wherein, yhkIs the element of the h-th row and the k-th column in Y, YkminAnd ykmaxThe minimum and maximum values in the k column of Y,is yhkNormalized values;
step 2.3: obtaining mean square error sequence E ═ E by using trained neural network1,e2,...,eQ];
And step 3: clustering residual training based on an affine propagation clustering algorithm;
the clustering of the residual training based on the affine propagation clustering algorithm comprises the following processes:
step 3.1: randomly initializing an attraction matrix R, an attribution matrix A and a damping factor lambda between 0 and 1; updating an attraction degree matrix R and an attribution degree matrix A:
D=R+A
wherein i, k, i ', k' are indexes, and s (i, k) ═ | xi-xk||2Is similarity, D is decision matrix, R is attraction matrix, A is attribution matrix, lambda is damping factor,denotes the maximum value of A (i, k ') + s (i, k') in the ith row except for the kth column,indicates that i' can only take values in i and k, i.e., max {0, R (i, k) } + max {0, R (k, k) }, where max {0, R (i, k) } indicates the larger of 0 and R (i, k); in a corresponding manner, the first and second optical fibers are,meaning that i' can take the remaining values of 1 to Q except i and k;
step 3.2: finding out a point with a diagonal larger than 0 in the decision matrix D as a clustering center, and classifying the sample points according to the distance between the sample points and the clustering representative points;
step 3.3: if the iteration times are larger than the initialized maximum iteration times, stopping clustering; otherwise, returning to the step 3.2;
and 4, step 4: and judging abnormal data according to the characteristics of each category and the intra-category distance.
2. The method for identifying abnormal data of an electric power system according to claim 1, wherein: the method for training the neural network by taking normal data of the power system as a training sample comprises the following steps:
step 1.1: acquiring a state measurement value data set X in the normal electric appliance operation process:
wherein, each row in X is a state measurement value of the running process of the electric appliance, in the matrix, M measurement values are counted, and each measurement value has N sampling data;
step 1.2: normalizing X:
wherein x isijIs the element in the ith row and the jth column in X, XjminAnd xjmaxIs the minimum of the jth column in XThe value and the maximum value of the sum,is xijNormalized values;
step 1.3: randomly splitting the state measurement value data set X along the row direction to form a set XM×LAnd XM×(N-L)Wherein L is N-L is approximately equal to 0.8:0.2 and is respectively used as a training set and a test set of the neural network;
step 1.4: setting the ith measurement value as a target value of neural network training, and setting the other measurement values as input values of the neural network training;
step 1.5: and training the neural network based on the particle swarm algorithm of the chaotic search.
3. The method for identifying abnormal data of an electric power system according to claim 2, wherein: the neural network is a three-layer neural network, the three layers comprise an input layer, a hidden layer and an output layer, the number of neurons of the input layer is M-1, the number of neurons of the hidden layer is set to be M-1, and the number of neurons of the output layer is 1;
the relationship between neural network inputs and outputs is:
where Y is the output value, f and g are both hyperbolic tangent activation functions, wijIs the input to the ith input layer neuron, αijIs the weight between the ith neuron of the input layer and the jth neuron of the hidden layer, βjIs the weight between the jth neuron of the hidden layer to the output node, AjThe bias coefficient of the jth neuron, the bias coefficient of the B output layer neuron, all weights and bias coefficients of the neural network are initialized randomly between-1 and 1; the error between the neural network output and the target value adopts the mean square error.
4. The method for identifying abnormal data of an electric power system according to claim 2, wherein: the chaotic search-based particle swarm algorithm for training the neural network comprises the following processes:
step 1.5.1: velocity v of randomly initialized particlesiAnd position li;
Step 1.5.2: calculating a fitness value f for each particleiI.e. the mean square error between the neural network output and the target value;
step 1.5.3: acquiring the historical optimal position of each particle and the optimal positions of all the particles;
step 1.5.4: updating speed and position;
step 1.5.5: judging whether the maximum iteration number is reached currently, if so, executing a step 1.5.6, otherwise, returning to the step 1.5.2;
step 1.5.6: and performing chaotic local search, and if the searched solution is more optimal, updating the historical optimal position until the mean square error between the neural network output on the test set and the target value is not reduced any more, wherein the obtained particle position is the optimization result.
5. The method for identifying abnormal data of an electric power system according to claim 4, wherein: the update speed and position are:
vid←w*vid+c1r1(Lid-lid)+c2r2(Lgd-lid)
lid←lid+vid
wherein: i is the particle index; d is the data dimension index, c1And c2Is a constant number r1And r2Is a closed interval [0,1]Random number of (3), LidFor the historical optimum position of the ith particle in the d-dimension, LgdFor the optimal position of all particles in the d-dimension, w is the inertial weight, lidAnd vidThe position and velocity of the ith particle in the d-dimension.
7. The method for identifying abnormal data of an electric power system according to claim 1, wherein: the judging abnormal data according to the characteristics of each category and the intra-category distance comprises the following steps:
if the number of the elements in a certain category is not more than 3, all the elements in the category are considered to be abnormal data;
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811609951.3A CN111382862B (en) | 2018-12-27 | 2018-12-27 | Method for identifying abnormal data of power system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811609951.3A CN111382862B (en) | 2018-12-27 | 2018-12-27 | Method for identifying abnormal data of power system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111382862A CN111382862A (en) | 2020-07-07 |
CN111382862B true CN111382862B (en) | 2021-09-14 |
Family
ID=71214490
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811609951.3A Active CN111382862B (en) | 2018-12-27 | 2018-12-27 | Method for identifying abnormal data of power system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111382862B (en) |
Families Citing this family (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112016248A (en) * | 2020-08-31 | 2020-12-01 | 华北电力大学 | Elman neural network based SCR denitration system bad data identification method |
CN112260989B (en) * | 2020-09-16 | 2021-07-30 | 湖南大学 | Power system and network malicious data attack detection method, system and storage medium |
CN113075498B (en) * | 2021-03-09 | 2022-05-20 | 华中科技大学 | Power distribution network traveling wave fault positioning method and system based on residual error clustering |
CN113421176B (en) * | 2021-07-16 | 2022-11-01 | 昆明学院 | Intelligent screening method for abnormal data in student score scores |
CN116881746B (en) * | 2023-09-08 | 2023-11-14 | 国网江苏省电力有限公司常州供电分公司 | Identification method and identification device for abnormal data in electric power system |
CN118152763A (en) * | 2024-05-11 | 2024-06-07 | 北京智芯微电子科技有限公司 | Distribution network data sampling method and device and electronic equipment |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104361393B (en) * | 2014-09-06 | 2018-02-27 | 华北电力大学 | Data predication method is used for based on the improved neural network model of particle swarm optimization algorithm |
US20160358075A1 (en) * | 2015-06-08 | 2016-12-08 | The Regents Of The University Of Michigan | System for implementing a sparse coding algorithm |
CN105224872B (en) * | 2015-09-30 | 2018-04-13 | 河南科技大学 | A kind of user's anomaly detection method based on neural network clustering |
CN108399201B (en) * | 2018-01-30 | 2020-05-12 | 武汉大学 | Web user access path prediction method based on recurrent neural network |
-
2018
- 2018-12-27 CN CN201811609951.3A patent/CN111382862B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN111382862A (en) | 2020-07-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111382862B (en) | Method for identifying abnormal data of power system | |
Fathi et al. | An improvement in RBF learning algorithm based on PSO for real time applications | |
CN110504676B (en) | Power distribution network state estimation method based on APSO-BP | |
CN111199252A (en) | Fault diagnosis method for intelligent operation and maintenance system of power communication network | |
CN111783845B (en) | Hidden false data injection attack detection method based on local linear embedding and extreme learning machine | |
CN111912611A (en) | Method and device for predicting fault state based on improved neural network | |
CN109901064B (en) | ICA-LVQ-based high-voltage circuit breaker fault diagnosis method | |
CN113723238A (en) | Human face lightweight network model construction method and human face recognition method | |
Bharill et al. | Enhanced cluster validity index for the evaluation of optimal number of clusters for Fuzzy C-Means algorithm | |
Xie et al. | Fuzzy multi-attribute decision making methods based on improved set pair analysis | |
Gu et al. | Efficient intrusion detection based on multiple neural network classifiers with improved genetic algorithm. | |
CN113255880A (en) | Method and system for judging electricity stealing data based on improved neural network model | |
Purwar et al. | Empirical evaluation of algorithms to impute missing values for financial dataset | |
Lin et al. | Some continuous aggregation operators with interval-valued intuitionistic fuzzy information and their application to decision making | |
CN108446506B (en) | Uncertain system modeling method based on interval feedback neural network | |
CN116680969A (en) | Filler evaluation parameter prediction method and device for PSO-BP algorithm | |
Zhao et al. | An effective model selection criterion for mixtures of Gaussian processes | |
Chen et al. | A GPU-accelerated approximate algorithm for incremental learning of Gaussian mixture model | |
Soltani et al. | A pso-based fuzzy c-regression model applied to nonlinear data modeling | |
CN114692729A (en) | New energy station bad data identification and correction method based on deep learning | |
CN114169007A (en) | Medical privacy data identification method based on dynamic neural network | |
Subbarayan et al. | Modular neural network architecture using piece-wise linear mapping | |
CN113707213A (en) | Protein-ligand binding site prediction method based on deep learning | |
Jun et al. | A multi-attribute group decision-making approach based on uncertain linguistic information | |
Ning | Network intrusion classification based on probabilistic neural network |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |